如何在R中使用Keras从深度学习中获得平衡的准确性？

install.packages("keras")
install.packages("tensorflow")

library(keras)

set.seed(123)

data <- iris  # 示例使用鸢尾花数据集

# 示例：对数据进行归一化
data <- scale(data)

# 示例：将数据集划分为训练集和测试集，比例为70%训练集，30%测试集
train_indices <- sample(1:nrow(data), 0.7*nrow(data))
train_data <- data[train_indices, ]
test_data <- data[-train_indices, ]

model <- keras_model_sequential()
model %>%
  layer_dense(units = 16, activation = "relu", input_shape = ncol(train_data)) %>%
  layer_dense(units = 3, activation = "softmax")

model %>% compile(
  loss = "categorical_crossentropy",
  optimizer = "adam",
  metrics = c("accuracy")
)

model %>% fit(
  x = as.matrix(train_data[, -ncol(train_data)]),
  y = to_categorical(train_data[, ncol(train_data)]),
  epochs = 10,
  batch_size = 32
)

model %>% evaluate(
  x = as.matrix(test_data[, -ncol(test_data)]),
  y = to_categorical(test_data[, ncol(test_data)])
)

predictions <- model %>% predict_classes(as.matrix(test_data[, -ncol(test_data)]))